Depth profile of the ferromagnetic order in a YBa${}_2$Cu${}_3$O${}_7$/La${}_{2/3}$Ca${}_{1/3}$MnO${}_3$ superlattice on a LSAT substrate: A polarized neutron reflectometry study

Using polarized neutron reflectometry (PNR) we have investigated a [${\mathrm{YBa}}_2{\mathrm{Cu}}_3{\mathrm{O}}_7$(10 nm)/${\mathrm{La}}_{2/3}{\mathrm{Ca}}_{1/3}{\mathrm{MnO}}_3$(9 nm)]${}_{10}$ (YBCO/LCMO) superlattice grown by pulsed laser deposition on a ${\mathrm{La}}_{0.3}{\mathrm{Sr}}_{0.7}{\mathrm{Al}}_{0.65}{\mathrm{Ta}}_{0.35}{\mathrm{O}}_3$ (LSAT) substrate. Due to the high structural quality of the superlattice and the substrate, the specular reflectivity signal extends with a high signal-to-background ratio beyond the fourth-order superlattice Bragg peak. This allows us to obtain more detailed and reliable information about the magnetic depth profile than in previous PNR studies on similar superlattices that were partially impeded by problems related to the low-temperature structural transitions of the SrTiO${}_3$ substrates. In agreement with the previous reports, our PNR data reveal a strong magnetic proximity effect showing that the depth profile of the magnetic potential differs significantly from the one of the nuclear potential that is given by the YBCO and LCMO layer thickness. We present fits of the PNR data using different simple blocklike models for which either a large ferromagnetic moment is induced on the YBCO side of the interfaces or the ferromagnetic order is suppressed on the LCMO side. We show that a good agreement with the PNR data and with the average magnetization as obtained from dc magnetization data can only be obtained with the latter model where a so-called depleted layer with a strongly suppressed ferromagnetic moment develops on the LCMO side of the interfaces. We also show that the PNR data are still compatible with the presence of a small, ferromagnetic Cu moment of $0.25{\mu }_{\mathrm{B}}$ on the YBCO side that was previously identified with x-ray magnetic circular dichroism and x-ray resonant magnetic reflectometry measurements on the same superlattice [D. K. Satapathy et al., Phys. Rev. Lett. 108, 197201 (2012)]. We discuss that the depleted layer thus should not be mistaken with a “dead” layer that is entirely nonmagnetic but rather may contain a canted antiferromagnetic or an oscillatory type of ordering of the Mn moments that is not detected with the PNR technique.

Figure 1

Temperature dependence of the resistance and the magnetic moment in field-cooled (FC) mode with $H=100$ Oe applied parallel to the layers as measured on the YBCO/LCMO superlattice. It shows the onset of the superconducting transition at $T_{\mathrm{c}}=88$ K and the ferromagnetic transition at $T^{\mathrm{Curie}}=201$ K. Inset: Magnetic hysteresis loops measured at 10 K and 100 K.

Figure 2

Unpolarized neutron reflectivity curves of the YBCO/LCMO SL measured at room temperature with the NREX and SuperADAM instruments. The curves are vertically shifted for clarity. Symbols show the experimental data and solid lines the best fits that were obtained by fitting both curves simultaneously. The arrows mark the position of the SLBPs. Inset: Symbols show the nonresonant XRR curve at 300 K. The solid line shows a simulation using the parameters as obtained from the fits of the neutron reflectometry curves.

Figure 3

(a) Polarized neutron reflectivity curves of the YBCO/LCMO SL measured at low temperature after field cooling in 100 Oe at SuperADAM for up $|+\rangle$ and down $|-\rangle$ polarization of the neutron spin with respect to the direction of the applied magnetic field. The lines show the best fit for the depleted layer model for $|+\rangle$ (dashed) and $|-\rangle$ (solid) neutron spin channels. For clarity the curves at 10 K are vertically shifted down by a factor of ${10}^2$. (b) Close-up on a linear intensity scale in the region of the second and third SLBPs to aid the comparison with the fit. The depth profiles of the used nuclear and magnetic scattering length densities are shown in (c). The same data are shown in (d)–(f) together with the best fit using the model of an inverse magnetic proximity effect and, in (g)–(i), for the model of an induced FM moment in YBCO that is parallel to the one in LCMO.

Figure 4

Same as in Fig. 3 but for the PNR data measured at NREX with an applied field of 4 kOe.

Figure 5

Maps of the off-specular reflection of the YBCO/LCMO superlattice measured (a) with unpolarized neutrons at 300 K and (b) for the $|+\rangle$ spin channel at 4 K after field cooling in a field of 100 Oe.

Figure 6

Room-temperature reflectivity curves (symbols) of the YBCO/LCMO SL measured with (a) neutrons and (b) x rays. The dashed lines show the best fit with a roughness of 8.5 Å at both the top and bottom interfaces that was also shown in Fig. 2. The solid lines show a simulation in which a roughness of 16 Å and 9.5 Å was assumed for the LCMO bottom and top interfaces, respectively. Insets: Magnification around the high-order SLBPs to highlight the difference between the two models.

Figure 7

Low-temperature PNR curves of the YBCO/LCMO SL at 4 kOe. The solid and dashed lines are the results of the fit using model 1a; here the magnetization in the depleted layers is fitted and the thicknesses of the top and bottom interfaces are set as common among all data sets.

Figure 8

Comparison of the average magnetic moment as determined experimentally from field-cooled dc magnetization measurements at 100 Oe and 4 kOe (solid lines) and calculated from the magnetic potential obtained with model 1 (circles), model 1a (squares), model 2 (upwards triangles), and model 3 (downwards triangles) from the fits to the PNR curves measured at 100 Oe (open symbols) and 4 kOe (solid symbols). The size of the symbols of the calculated magnetic moments reflects the error bars. The error bars of the dc magnetization data arise from the statistical errors and the uncertainty of the volume of the small piece used for the dc magnetization measurements.